Estimating and predicting the operational costs of road tolls: An econometric assessment using panel data. (December 2019)

• Record Type: Journal Article
• Authors: Odeck, James
• Is Part Of: Transportation research. Volume 130(2019)
• Page Start: 466
• Page End: 478
• Publication Date: 2019-12
• Subjects: Road tolls -- Estimating operational cost -- Predictions -- Econometrics -- Norway; Transportation -- Research -- Periodicals; 388.011
• DOI: 10.1016/j.tra.2019.09.047
• Languages: English
• ISSNs: 0965-8564

Abstract

Abstract: We develop an econometric procedure to estimate the operational costs of toll road projects. To show its usefulness for planning future tolls, we test its predictive power on out-of-the sample observations. Thus, the rationale for the paper is that previous studies have utilized less succinct econometric frameworks to derive factors that significantly explain the operational costs of toll roads and have provided little evidence regarding the predictive power of the models used. Panel data on 33 Norwegian toll road projects observed during the period 2011–2015 are used for the purpose, resulting in a total of 165 observations. The results reveal the following: (i) Toll operations are heterogeneous, so that individual characteristics matter; ceteris paribus, this should be accounted for when estimating operational costs. (ii) The average operational cost per vehicle (AVC) processed at tolls is approximately 4 NOK (≈ US$ 0.44). (iii) The volume of vehicle throughput reduces operational costs, demonstrating the existence of scale economies; the number of lanes served increases operational costs; the use of an Open Road Toll (ORT) charging system reduces operational costs; the tendering of toll collections reduces operational costs; and the share of vehicles that use On Board Units (OBU) reduces operational costs. It is shown that the econometric model developed here has high predictive power and can be used to predict the operational costs of future tolls. These results reaffirm the results of some earlier studies but serve to warn that one should not be indifferent to the framework of analysis used, as the results may not be statistically significant and robust. The econometric framework developed here is an advance due to its predictive power.